How do I replace texts based on conditions on multiple other dataframe in python?

I apologize in advance for a long winded question. I am new to python and not really sure where to begin with this problem.

I have a column 'car_number' in df that sometimes has UNKNOWN values. When there is an UNKNOWN:

• Use the 'delivery' date in that row and check it against the oldest 'delivery' date of the same 'car_part' number in df2
• If the dates matches within a 3 months period for df and df2 forward or backward
  • Go to df3 and match 'delivery' date using same 3 months condition forward or backward
    • If it matches, then take the lowest car_number and replace it with UNKNOWN

Once UNKNOWN is replaced, create a new column called formula in df and applies a formula to check if the date will return a negative. The formula:

• Checks if the new 'car_number' matches the 'car_number' on the next row.
• If it matches then take 'delivery' date and subtract 'test' date and returns number of days between the two dates.
• If the number is positive or blank, then moves onto the next UNKNOWN
• If the number is negative
  • Then go back to df3 and pick the next car_number and replace
  • Applies the formula again until a positive number is returned
• If the new car_number doesn't match the car_number on the next row, returns nothing.

The end result should show that the first UNKNOWN is replaced by X123-00027C and second UNKNOWN by X123-00148C. Again, sorry for the long question.

data = {'car_part': ['100009','100093','100071','100033','100033','100043'],
        'car_number': ['UNKNOWN', 'X123-00027C', 'X123-00027C', 'UNKNOWN', 'X123-00148C', 'X123-00148C'],
        'delivery': ['11/20/2004', '12/17/2009', '7/27/2010', '11/1/2004', '9/5/2004', '11/10/2004'],
        'test': ['12/17/2009', '7/27/2010', '7/10/2020', '12/22/2006', '3/26/2007', '12/1/2007']}  

data2 = {'delivery': ['11/1/2004', '12/1/2004', '1/1/2005', '7/1/2006', '8/1/2006', '9/2/2006'], 
         'car_part': ['100009','100009','100009','100033','100033','100033']}  

data3 = {'car_part': ['100009','100009','100009','100033','100033','100033'],
         'car_number': ['X123-00027C', 'X123-00031C', 'X123-00041C', 'X123-00106C', 'X123-00143C', 'X123-00148C'],
         'delivery': ['8/27/2004', '8/27/2004', '8/27/2004', '3/27/2006', '3/27/2006', '3/27/2006']}  

df = pd.DataFrame(data)
print(df)
df2 = pd.DataFrame(data2)
print(df2)
df3 = pd.DataFrame(data3)
print(df3)

df['delivery'] = df['delivery'].astype('datetime64[ns]')
df.sort_values(by = ['car_part', 'delivery', 'test'], ascending=[True, True, True])

df2['delivery'] = df2['delivery'].astype('datetime64[ns]')
df2.sort_values(by = ['car_part', 'delivery'], ascending=[True, True])

df3['delivery'] = df3['delivery'].astype('datetime64[ns]')
df3.sort_values(by = ['car_part', 'car_number', 'delivery'], ascending=[True, True, True])